Atmospheric Temperature And Humidity Profiles Retrieval From Hyperspectral Infrared Simulation Data Based On FY-4

Atmospheric temperature and humidity are the key parameters of meteorological forecast and climate prediction, and hyperspectral Infrared remote sensing provide regional and global high-precision temperature and humidity. There is no relevant research and retrieval of same track, and the demand of retrieval temperature and humidity is more and more urgent.In order to promote the research of retrieving temperature and humidity based on FY-4 GIIRS, we use neural network to construct four retrieval systems: retrieval algorithm models of temperature and humidity on global and Chinese region. At the same time, we also build retrieval systems of temperature and humidity based on IASI, and compare the retrieval capacity about the two kinds of data.The main research results and conclusions of this paper are shown as follows:(1) This paper successfully complete the forward simulation of GIIRS and IASI.Using CIMSS as the input of radiative transfer model to calculate TB of GIIRS and IASI, and comparing the bright temperatures to find that although simulative TB of GIIRS is lower than IASI, the trend of both at all channels is consistent, which verifies the accuracy of forward simulation on GIIRS.(2) This paper successfully complete the retrieval of temperature base on simulative data of GIIRS. Two retrieval models of temperature are eatablished in the world and China. It's found that retrieval accuracy of temperature in China(RMSE=1.922k) is higher than in the world (RMSE=2.630K). At the same time, the effect of retrieval temperature on five kinds of CIMSS data is compared and analyzed. The retrieval accuracy of temperature on NOAA88 is the highest(RMSE=1.853K), the average error reaches 0.007K, and has high correlation(R=0.998). The retrieval precision of troposphere is obviously higher than that in the stratosphere. The root mean square errors (RMSE) of retrieval atmospheric temperature in the troposphere and stratosphere is 0.846K and 2.020K, and the average errors are -0.003K and 0.024K , respectively.(3) This paper successfully complete the retrieval of humidity base on simulative data of GIIRS.Two retrieval models of temperature are established in the world and China. For the retrieval models in the world, the RMSE of water vapor mixing ratio is 0.440g/kg. What's more, it's found that the retrieval accuracy of humidity on NOAA88 is the highest (RMSE=0.315g/kg). For the retrieval model in China, the RMSE of minxing ratio, relative humidity and water vapor density reach 0.362g/kg, 9.079% and 0.450g/m3, respectively. At the same time, the model accuracy of retrieval humidity in China is better than in the world.(4) This paper successfully complete the retrieval of temperature and humidity base on simulative data of IASI. For the retrieval model in the world, the RMSE of temperature and humidity reaches 1.054K and 0.524g/kg, and we find that the retrieval accuracy of humidity on NOAA88 is the highest (RMSE=0.439g/kg). For the retrieval model in China, the RMSE of temperature, minxing ratio, relative humidity and water vapor density respectively reach 0.989K, 0.487g/kg, 5.307% and 0.364g/m3, respectively. Finally, it is found that the retrieval of temperature and humidity in China is better than in the world.(5) The accuracy of retrieval temperature and humidity based on GIIRS and IASI is compared. Inspecting the capacity of retrieval temperature on simulative data,it is found that the accuracy of retrieval on GIRRS is better than IASI at the bottom,but worse than IASI on top of 500hPa. What's more, inspecting the capacity of retrieval humidity on simulative data, we find that the accuracy of retrieval on GIIRS is better than IASI at all layers. In short, the data of GIIRS based on FY-4 has a great research prospect in the retrieval of atmospheric temperature and humidity profiles.